Deployment fails because of wrong option stats in haproxy config for some backends

I am attempting to deploy to three nodes in the following environment.
Fuel master and the three nodes are running as VMs in a vsphere cluster with vsan storage, DRS is fully automated and I am using a VDS. Networking checks out fine.
I'm deploying the OS instance with CentOS.
The three nodes are controller/cinder proxy for vcenter.
No compute or storage node. QEMU + vCenter.
No ceph
Glance configured for vcenter datastores.

The first node appears to deploy fine.
The second two nodes fail on puppet task for verifying haproxy mysqld.
haproxy and mysqld are running on both of these nodes, when I try to restart haproxy I get
[WARNING] 071/154537 (25447) : config : 'stats' statement ignored for proxy 'mysqld' as it requires HTTP mode.

This is reported in bug #1431933
However, in the bug report the stats ignored because it requires HTTP mode appears for multiple services. In my case it appears only for mysqld.

I did notice that nova reports no storage from they hypervisor stats. In the vsphere environment I am running vsanDatastore. In compute service I used the regex vsan* and for glance I used datacenter = CTI-Datacenter, datastore = vsanDatastore

One other thing I noticed which is probably unrelated is that one of the logs (I'm sorry I don't remember which) reports no cache, assuming write through cache as an error. This appears on the node that is running correctly so I assume it is not causing an issue and may be related to the fact that I have no pre-allocated any flash cache to the VM's.

Another thing I took note of was that when the fuel install finished the UI told me their was updates but yum reports 0 updates. yum repolist reports that the mos repos have no packages. There is not a networking issued on the fuel vm.

I am currently trying a deployment on ubuntu but I won't be able to check the success of it until tomorrow.

I was hoping someone could tell me how to work around this issue since it is marked as not being fixed in 6.1. I am doing something wrong in the deployment configuration? Would deploying 6.0 have the same issue? Should I deploy 1 controller node and then add the other two? etc.